A tutorial that consolidates the theory and practice of correlated noise (factorization and matrix) mechanisms for differentially private optimization and prefix sum estimation, without introducing a new central result.
Optimality of Matrix Mechanism on $\ell_p^p$-metric
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
In this paper, we introduce the $\ell_p^p$-error metric (for $p \geq 2$) when answering linear queries under the constraint of differential privacy. We characterize such an error under $(\epsilon,\delta)$-differential privacy. Before this paper, tight characterization in the hardness of privately answering linear queries was known under $\ell_2^2$-error metric (Edmonds et al., STOC 2020) and $\ell_p^2$-error metric for unbiased mechanisms (Nikolov and Tang, ITCS 2024). As a direct consequence of our results, we give tight bounds on answering prefix sum and parity queries under differential privacy for all constant $p$ in terms of the $\ell_p^p$ error, generalizing the bounds in Henzinger et al. (SODA 2023) for $p=2$.
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Correlated Noise Mechanisms for Differentially Private Learning
A tutorial that consolidates the theory and practice of correlated noise (factorization and matrix) mechanisms for differentially private optimization and prefix sum estimation, without introducing a new central result.